ubic.basecode.math

Class CorrelationStats

java.lang.Object

ubic.basecode.math.CorrelationStats

public class CorrelationStats
extends Object

Statistical evaluation and transformation tools for correlations.

Author:

Paul Pavlidis

Constructor Summary

Constructors

Constructor and Description
CorrelationStats()

Method Summary

Modifier and Type	Method and Description
static double	byteToCorrel(int correlByte)
static double	byteToPvalue(int pvalByte)
static double	compare(double correl1, int n1, double correl2, int n2) Statistical comparison of two Pearson correlations.
static double	correl(double[] ival, double[] jval) Compute the Pearson correlation, missing values are permitted.
static int	correlAsByte(double correl)
static double	correlationForPvalue(double pval, int count) Find the approximate Pearson correlation required to meet a particular pvalue.
static double	correlationTstat(double correl, int dof) Compute the t-statistic associated with a Pearson correlation.
static double	correlFast(double[] ival, double[] jval, double meani, double meanj, double sqrti, double sqrtj) Compute Pearson correlation when there are no missing values.
static double	fisherTransform(double r) Compute the Fisher z transform of the Pearson correlation.
static cern.colt.list.DoubleArrayList	fisherTransform(cern.colt.list.DoubleArrayList e) Fisher-transform a list of Pearson correlations.
static boolean	isValidPearsonCorrelation(double r) Test if a value is a reasonable Pearson correlation (in the range -1 to 1; values outside of this range are acceptable within a small roundoff.
static double	pvalue(double correl, int count) Compute pvalue for the pearson correlation, using the t distribution method.
static int	pvalueAsByte(double correl, int count) Convert a p value into a value between 0 and 255 inclusive.
static double	spearmanPvalue(double correl, int count)
static double	unFisherTransform(double z) Reverse the Fisher z-transform of Pearson correlations

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

CorrelationStats

public CorrelationStats()

Method Detail

byteToCorrel

public static double byteToCorrel(int correlByte)

Parameters:

correlByte - int

Returns:

double

byteToPvalue

public static double byteToPvalue(int pvalByte)

Parameters:

pvalByte - int

Returns:

double

compare

public static double compare(double correl1,
                             int n1,
                             double correl2,
                             int n2)

Statistical comparison of two Pearson correlations. Assumes data are bivariate normal. Null hypothesis is that the two correlations are equal. See Zar (Biostatistics)

Parameters:

correl1 - First correlation

n1 - Number of values used to compute correl1

correl2 - Second correlation

n2 - Number of values used to compute correl2

Returns:

double p value.

correl

public static double correl(double[] ival,
                            double[] jval)

Compute the Pearson correlation, missing values are permitted.

Parameters:

ival -

jval -

Returns:

correlAsByte

public static int correlAsByte(double correl)

Parameters:

correl - double

Returns:

int

correlationForPvalue

public static double correlationForPvalue(double pval,
                                          int count)

Find the approximate Pearson correlation required to meet a particular pvalue. If the pvalue is <=0 or >= 1, returns 1 and 0 respectively.

Parameters:

pval - double

count - int

Returns:

double

correlationTstat

public static double correlationTstat(double correl,
                                      int dof)

Compute the t-statistic associated with a Pearson correlation.

Parameters:

correl - Pearson correlation

dof - Degrees of freedom (n - 2)

Returns:

double

correlFast

public static double correlFast(double[] ival,
                                double[] jval,
                                double meani,
                                double meanj,
                                double sqrti,
                                double sqrtj)

Compute Pearson correlation when there are no missing values.

Parameters:

ival -

jval -

meani -

meanj -

sqrti -

sqrtj -

Returns:

fisherTransform

public static double fisherTransform(double r)

Compute the Fisher z transform of the Pearson correlation.

Parameters:

r - Correlation coefficient.

Returns:

Fisher transform of the Correlation.

fisherTransform

public static cern.colt.list.DoubleArrayList fisherTransform(cern.colt.list.DoubleArrayList e)

Fisher-transform a list of Pearson correlations.

Parameters:

e -

Returns:

isValidPearsonCorrelation

public static boolean isValidPearsonCorrelation(double r)

Test if a value is a reasonable Pearson correlation (in the range -1 to 1; values outside of this range are acceptable within a small roundoff.

Parameters:

r -

Returns:

pvalue

public static double pvalue(double correl,
                            int count)

Compute pvalue for the pearson correlation, using the t distribution method.

Parameters:

correl - Pearson correlation.

count - Number of items used to calculate the correlation. NOT the degrees of freedom.

Returns:

double one-side pvalue

pvalueAsByte

public static int pvalueAsByte(double correl,
                               int count)

Convert a p value into a value between 0 and 255 inclusive. This is done by taking the log, multiplying it by a fixed value (currently 8). This means that pvalues less than 10^-32 are rounded to 10^-32.

Parameters:

correl - double

count - int

Returns:

int

spearmanPvalue

public static double spearmanPvalue(double correl,
                                    int count)

Parameters:

correl - Spearman's correlation

count -

Returns:

pvalue, two-tailed pvalue, computed using t distribution for large sample sizes or exact computation for small sample sizes.

unFisherTransform

public static double unFisherTransform(double z)

Reverse the Fisher z-transform of Pearson correlations

Parameters:

z -

Returns:

r